art 4

download art 4

of 23

description

4

Transcript of art 4

  • ed Inclusive Innovation

    oping World?

    nd

    ,e C

    acronnoider knrogleveof qtmeriesof

    rnsanrode spf thtion

    2015 Elsevier Ltd. All rights reserved.

    Thecountrisets ofeasily daccesstion ariedge (benetSuch kthey geKrugmsupporoften v

    oine knowledge and consequently may have less to gain

    ity and innovation performance. It analyses whether the Inter-net as a conduit of knowledge spillovers has heterogeneous

    erging

    indus-catingnowl-oduc-ent inatentsf rmts. Wetions.ces in

    to bevation

    for valuable comments. Valentina Rollo gratefully acknowledges support

    from the SNF, Project Number PDFMP1_135148. The ndings expressed

    in this paper are those of the authors and do not necessarily represent the

    World Development Vol. 78, pp. 587609, 20160305-750X/ 2015 Elsevier Ltd. All rights reserved.

    6/j.worlddev.2015.10.029impacts on groups of rms that dier with regard to theirfrom Internet-enabled knowledge transfers. However, lessinnovative rms may lack capabilities to use newly availableknowledge for their business purposes (Cohen & Levinthal,1989).This paper provides evidence of knowledge spillover eects

    from industries adoption of the Internet on rms productiv-

    *The authors would like to thank Richard Baldwin, Nicolas Berman, Ana

    Margarida Fernandes, Dominique Guellec, Eric J. Bartelsman and

    participants of the 2014 ABCDE Conference, the Inter-American

    Development Banks internal seminar series and the 7th Annual

    Conference of the Academy of Innovation and Entrepreneurship (CAED)developing economies (OECD, 2015; Paunov, 2013). TheInternet may help groups of rms that engage less ininnovation, by improving their access to knowledge. Bycontrast, leading innovators often have access to quality

    performance and, hence, the risk of reverse causality is low.Notwithstanding, we cannot exclude a small risk of endogene-ity which would arise if the most productive and innovativerms use of the Internet caused other rms in the industryinnovacompreKey words Information and communication technology (ICT), knowledge spillovers, Internet, innovation, productivity, rm heter-ogeneities

    1. INTRODUCTION

    adoption of the Internet has been widespread acrosses (ITU, 2014). With the Internet, increasingly largeknowledge and information (big data) can be moreiused to large groups of people. By allowing for widerto ideas, the Internet may boost innovation as innova-ses from new combinations of existing pieces of knowl-Arthur, 2007). With the Internet, opportunities tofrom knowledge created by others are possibly higher.nowledge spillovers are critical for economic growth asnerate increasing returns (Grossman & Helpman, 1991;an, 1991; Romer, 1986). In addition, the Internet cant more inclusive innovation, i.e., the widening of theery small group of innovating rms in emerging and

    50,013 rm observations for 117 developing and emcountries for the 200611 period.Our empirical methodology exploits information on

    tries adoption of the Internet as a tool for communiwith suppliers and clients to identify Internet-enabled kedge spillover eects. Our specication regresses rm prtivity and innovation performance i.e., their investmequipment and ownership of quality certicates and p on industries use of the Internet, a comprehensive set ocontrols as well as industry and country-year xed eeccontrol for rms own use of the Internet in all specicaOur identication exploits within country-year dierenthe adoption of the Internet across industries.An industrys adoption of the Internet is unlikely

    aected by an individual rms productivity and innoHas the Internet Foster

    in the Devel

    CAROLINE PAUNOVa aaOECD

    b International Trad

    Summary. The adoption of the Internet has been widespreadfacilitating knowledge diusion among businesses to boost their iuse this newly available knowledge could create a new digital divand emerging countries over the 200611 period, this paper tests foon rms productivity and innovation performance. We test for heteengaged in innovation and on rms with dierent productivityperformance i.e., their investment in equipment and ownershipSpillover eects are identied by controlling for rms own invesas well as extensive rm-level controls. Our results show that industand its investment in equipment. We also identify modest impactsquality certicates and patents. On average, we nd that the retuinnovation, including single-plant establishments, non-exporters,quantile regressions show that only the most productive rms reap pproductivity levels below the 50th percentile do not benet much. Thour work justify public policies aimed at fostering industries use oabsorptive capabilities benet from the widespread Internet adopnetworks and strengthening their capacities to use them.

    www.elsevier.com/locate/worlddevhttp://dx.doi.org/10.101tion performance and productivity. This study provideshensive evidence on these questions for a sample of

    587VALENTINA ROLLOb,*

    Franceentre, Switzerland

    ss countries, making much more information available and thusvation performance. However, dierences in rms capabilities toinstead. Using 50,013 rm observations covering 117 developingowledge spillover eects from industries adoption of the Interneteneous spillover impacts on groups of rms that are commonly lessls. Our specication regresses rm productivity and innovationuality certicates and patents on industries use of the Internet.nt in Internet technology, industry and country-year xed eectsuse of the Internet positively aects the average rms productivityindustries use of the Internet on the likelihood that rms obtainto productivity are larger for rms that commonly engage less ind rms located in small agglomerations. However, results fromuctivity gains from Internet-enabled knowledge access. Firms withillover eects from industries adoption of the Internet identied ine Internet. However, since we show that only rms with adequate, policy support should also focus on facilitating rms access toviews of the OECD, the International Trade Center or their member

    countries. Final revision accepted: October 3, 2015.

  • spillovers. Several studies have shown that gross social returnsto knowledge investments exceed private returns (Bloom,

    VEto adopt the Internet. Consequently, higher industry adoptionrates could be positively correlated with strong performance.We address these endogeneity concerns as part of our robust-ness tests.We use ordinary least squares regressions to study aggregate

    eects on rm productivity and equipment investment rates aswell as logistic and probit regressions to analyze impacts onrms ownership of quality certicates and patents. In addi-tion, we apply quantile regressions to test whether benetsfrom industries adoption of the Internet dier across rmsof dierent productivity levels, which proxy for their capacitiesto absorb new knowledge. We also use quantile regression toensure our results are not driven by non-normal errors andoutliers as might be the case for ordinary least squares regres-sions.We nd that industries use of the Internet has positive

    impacts on rms labor productivity and on their investmentsin equipment. We also identify modest impacts on the likeli-hood that rms obtain quality certicates and patents. Theevidence is robust to various tests such as removing potentialoutliers and using alternative sources of Internet-enabledknowledge spillovers, including the Internet uptake by rmswithin geographic locations and industries as well as at thecountry rather than at the country-industry level.Moreover, we show that, on average, the Internet adoption

    of their industries benets more groups of rms that com-monly engage less in innovation: rms that do not export,rms that are not part of multi-plant establishments and rmsthat operate in small agglomerations.Quantile regression results show that the more productive

    rms gain more from their industries intensive use of theInternet. Firms with productivity levels below the 50th per-centile do not benet much. Moreover, only the most produc-tive non-exporting rms and single-plant establishmentsbenet from knowledge spillovers. Interestingly, while we donot nd that rm size aects productivity gains of the averagerm, quantile regression results reveal larger payos for themost productive small rms compared to larger rms.Several policy implications arise from our analysis. First, the

    existence of spillover eects from industries adoption of theInternet, which do not depend on rms own investments, jus-tify public policies aimed at fostering industries use of theInternet. The potential returns from policy support of indus-tries Internet adoption are high because, dierently fromother private sector development policies, benets arise evenfor rms that commonly engage less in innovation and forrms that face cumbersome business environment conditions(Paunov & Rollo, 2015).However, dierences in capabilities to use the knowledge

    made available on the Internet could create a new digitaldivide. Only rms with absorptive capabilities can benetfrom business intelligence platforms, which give access toknowledge relevant to their scientic and technological needs.Therefore, facilitating rms access to such networks andstrengthening their capacities to use them deserve policy sup-port. Human capital investments are core complementarypolicies (Indjikian & Siegel, 2005).This paper relates to the research on the contributions of

    ICT to development. An ongoing debate focuses on adequateinfrastructure conditions in developing and emerging econo-mies such as bandwidth capacity and aordable access prices.Forbiddingly high prices to access the Internet, which may becaused by technical or market imperfections, can reduceuptake (Hilbert, 2010; Howard & Mazaheri, 2009). These

    588 WORLD DEare pre-conditions for rms to use ICT for their business oper-ations to support their productivity (Ref. Section 2(a)).Schankerman, & Van Reenen, 2013). Audretsch andFeldman (2004) and Keller (2004) discuss research ndingson the international and geographic dimensions of knowledgespillovers. There is also evidence on barriers to knowledge spil-lovers, including from foreign direct investment (see Gorg &Greenaway, 2004). Firms lack of absorptive capacity tomake use of newly available knowledge explains some of thelimitations (Girma, 2005; Kokko, 1994; Kokko, Tansini, &Zejan, 1996). Geographical proximity also matters for spil-lovers (Ref. Section 2(c)).Our paper makes several contributions to the literature. To

    the best of our knowledge, this is the rst study to providecomprehensive cross-country evidence of knowledge spilloversof industries Internet adoption on rm productivity and inno-vation performance. Using data for the 200611 period allowsestimating global impacts at a point in time when the Internetadoption gained some maturity. Data for earlier years mayunderestimate impacts. Moreover, our study expands on theprevious analyses by testing whether groups of rms that com-monly innovate less benet more from their industries Inter-net adoption. In addition, we apply quantile regressiontechniques to explore whether average eects obtained fromconventional estimation techniques hide dierences inimpacts across rms of dierent productivity levels.The remainder of the paper is organized as follows. Section 2

    discusses the conceptual framework while Section 3 presentsthe data we use for our analysis. Section 4 introduces theempirical framework. Section 5 provides descriptive statisticswhile Section 6 describes the results of the analysis. Section 7concludes.

    2. CONCEPTUAL FRAMEWORK

    (a) ICT investments as driver of eciency improvements

    Firms, industries, and countries that invested in ICT haveimproved the eciency in which they transform inputs into out-puts. Research, conducted at industry and rm levels, has con-sistently found that ICT investments positively relate to higherproductivity (e.g., Bartel, Ichniowski, & Shaw, 2007; Bloom,Sadun, & Van Reenen, 2012; Jorgenson, 2001; Jorgenson &Vu, 2005; Oliner & Sichel, 2000; Stiroh, 2002). 2 There is alsosome evidence for rms in developing and emerging economies(e.g., Commander, Harrison, &Menezes-Filho, 2011; ECLAC,2011;Motohashi, 2008; Pohjola, 2001;UNCTAD, 2008;WorldBank, 2006). In addition, rms ICT investments also relateMoreover, our work relates to studies on the opportunitiesfor ICT to support very small rms and entrepreneurs fromdisadvantaged socio-economic backgrounds. ICT have helpedsmallholder farmers and sheries obtain market information(including on price trends) as well as knowledge about produc-tion techniques (Jensen, 2007; Muto & Yamano, 2009; Ogutu,Okello, & Otieno, 2014). Several studies have identied gainsfrom ICT-based services for disadvantaged producers (e.g.,Aker & Mbiti, 2010; Donner, 2004, 2006; Donner &Escobari, 2010; Duncombe & Heeks, 2002; Esselaar, Stork,Ndiwalana, & Deen-Swarra, 2007; Kaushik & Singh, 2004).However, improved access to knowledge has not always ben-etted these groups as they often lack capabilities to exploitnew knowledge (e.g., Tadesse & Bahiigwa, 2015). 1

    In addition, our paper relates to the literature on knowledge

    LOPMENTpositively to their innovation performance (see, for example,Spezia, 2011, for an analysis of eight OECD countries).

  • INHowever, ICT do not automatically boost productivity. Evi-dence prior to the mid-1990s did not identify such positiveimpacts (Brynjolfsson & Yang, 1996). Hence, Solows famousquote (1987) you can see the computer age everywhere but inthe productivity statistics. Limited productivity gains alsoreected the need for production process adjustments beforeICT had positive productivity impacts, including the needfor organizational changes and adequate human capital.The complementarity between on the one hand ICT invest-

    ment and on the other hand organizational change and man-agement capacities is critical for ICT investments to boostproductivity (Black & Lynch, 2001, 2004; Bloom et al., 2012;Bresnahan, Brynjolfsson, & Hitt, 2002; Bresnahan,Greenstein, Brownstone, & Flamm, 1996). Dierences in orga-nizational and managerial capabilities are among the reasonswhy ICT contributed more to US than to Europes productiv-ity (Cardona et al., 2013). Bloom et al. (2012) nd that forrms operating in the United Kingdom, the productivity ofICT capital has been signicantly higher in US-owned estab-lishments compared to other rms.

    (b) Knowledge spillovers and rms innovation performance

    In addition to rms productivity improvements from theirown ICT investments, further productivity and innovationperformance gains may arise from industries adoption ofthe Internet as a means of communication to improve the dif-fusion of knowledge. The Internets contribution is well illus-trated by an analogy: drawing balls from an urn that holdsrelevant knowledge. The Internet helps access a larger numberof balls from that urn. Improved communication among mem-bers of an industry facilitates learning about new technologiesand consequently accelerates the rate of innovation (e.g.,Conley & Udry, 2010 and references therein).Information from clients, suppliers, and competitors can

    strengthen rms innovation performance in the followingways: First, information on customer preferences helps iden-tify market opportunities for new products and services.Uncertainty about future market demands for new productsis a major obstacle for rms to invest in innovation(Collard-Wexler, Asker, & De Loecker, 2011). User feedbackcan also help rms develop new product and services; it is usedsystematically to identify bugs in software codes and to createimproved software programs. 3 Second, developments of tech-nology relevant to rms production processes determine thetechnical feasibility of new products and processes. Hence,better communication with suppliers to learn about new tech-nological possibilities and discuss rms needs can supportinnovation. Third, knowledge about competitors practicesallows rms to learn about alternative production techniquesand innovations. However, knowledge from competitorsmight be less accessible because they have an interest in keep-ing information secret from each other (see e.g., Fernandes &Paunov, 2012; Javorcik, 2004).Aside from knowledge spillovers, there are other potential

    sources of benet from industries adoption of the Interneton rms productivity and innovation performance. Wide-spread adoption of the Internet allows rms to order onlinefrom suppliers and deliver products more eciently to cus-tomers. The use of ICT can also improve the evidence-basein rms decision-making (e.g., Brynjolfsson, Hitt, & Kim,2011). Taken together these impacts are closely related torms own adoption of the Internet rather than to their indus-tries adoption. Knowledge spillovers are likely to be more

    HAS THE INTERNET FOSTERED INCLUSIVEimportant sources of productivity and innovation perfor-mance gains from industries Internet adoption.(c) Knowledge spillovers and the Internet

    Knowledge lends itself to spillovers since, once created, itcan be replicated and disseminated at virtually no cost, andconsequently benet more rms (Arrow, 1962). The Internethas contributed to reducing dissemination costs further. How-ever, there are barriers to Internet-enabled knowledge diu-sion: only codied knowledge can be transmitted while tacitknowledge cannot (Leamer & Storper, 2001). This is why geo-graphic proximity matters for knowledge diusion (Audretsch& Feldman, 1996; Krugman, 1991). However, ICT havereduced barriers for transmitting knowledge; for instance,videoconference opportunities mimic face-to-face interactionsbetter than other ways of communicating across geographicdistances.Potential benets from Internet-facilitated knowledge spil-

    lovers do not depend on the individual rms use of the Inter-net but on adoption by a critical mass of an industrys rms.Even rms that do not use the Internet can benet fromInternet-enabled knowledge diusion within their industryby other means, such as participation in business associationsand recruitment of sta from other rms. The questionwhether the Internet facilitates knowledge spillovers is, there-fore, distinct from asking how rms own adoption of ICTshas beneted their performance.

    (d) The Internet as a potential facilitator of inclusive innovation

    More inclusive innovation i.e., expanding the group of inno-vators to groups that are traditionally not engaged in innova-tion, is an important issue for developing and emergingeconomies. Small and young rms can be drivers of innova-tion as they often have greater agility to introduce novel ideas(Acs & Audretsch, 1990). Yet, fewer opportunities for theserms to access credit stie their contributions to innovation,particularly in developing and emerging economies. Largerincumbent businesses often have access to more fundingopportunities. Other obstacles are also less of a challenge forlarge incumbent rms compared to small and young rms.Consequently, a few incumbent rms are islands of excel-lence in a sea of smaller businesses of low productivity(OECD, 2015). Such industrial structures are inecient asaggregate productivity would be much higher if all rms wereas productive as the best performing ones (Hsieh & Klenow,2009). Wage inequality is also higher as less productive rmspay lower wages than the more productive rms (Mueller,Ouimet, & Simintzi, 2015). Potential Internet-enabled knowl-edge diusion to improve the performance of less productiverms is consequently critical.The benets from the Internet adoption of their industries

    may be larger for some rms than for others. Coming backto the analogy of the urn introduced above, better knowledgenetworks give rms access to all of the relevant knowledgesince the full number of available balls is xed. All else equal,rms that are connected to rich (poor) oine knowledge net-works may have fewer (stronger) productivity and innovationperformance gains from new online networks. Some groups ofrms dier in the quality of oine knowledge networks; thosewith bad quality network connections commonly also engageless in innovation. Dierences may arise for exporters andforeign-owned rms, rms located in large (small) agglomera-tions, rms of dierent sizes as well as multi- and single plantrms and informal businesses.First, exporters and foreign-ownedrmsmayhave less to gain

    NOVATION IN THE DEVELOPING WORLD? 589from Internet-enabled knowledge spillovers because they accessforeign expertise, which is a critical source of advanced

  • 3. DATA

    We use data from the second wave of the World BankEnterprise Surveys (WBES) for our empirical analysis. TheWBES collect information on a representative sample of for-mal rms in the countries non-agricultural sector; the selec-tion of rms is done by stratied random sampling (Dethier,Hirn, & Straub, 2011). The WBES have been widely used,including in Almeida and Fernandes (2008), Beck,Demirguc-Kunt, and Maksimovic (2008), Fisman andSvensson (2007) and Paunov (2016). 5

    Our analysis uses a cross-section of rms, information for50,013 rm observations across 117 countries for the 200611 period. This sample is drawn from 65,285 rm observationsavailable, excluding observations with missing information onlabor productivity and industries use of the Internet. 6 Table 1summarizes data coverage across world regions, industries,rm size categories, years, and country income levels.

    Table 1. Descriptive statistics

    Number ofobservations

    Share intotal (%)

    Region

    Africa 13,741 27.5Eastern Europe and Central Asia 9,968 19.9Latin America and the Caribbean 19,772 39.5Middle East 1,007 2.0East Asia Pacic 3,677 7.4South Asia 1,848 3.7

    Industry

    Food 6,326 12.7Garments 3,987 8.0Textiles and leather 2,567 5.1Wood and furniture 689 1.4Non-metallic and plastic materials 2,337 4.7Metals, machinery and electronics 3,738 7.5Chemicals and pharmaceuticals 2,387 4.8Other manufacturing activities 6,921 13.8Total manufacturing 28,952 57.9

    Services (incl. construction)

    Hotels and restaurants 1,816 3.6Retail and wholesale trade 11,641 23.3Construction and transportation 2,629 5.3Other services 4,975 10.0

    VELOPMENTtechnologies (Coe&Helpman, 1995; Fagerberg, 1994; Freeman& Soete, 1997). National rms and non-exporters may conse-quently benetmore from Internet-enabled knowledge transfer.Second, rms located in large agglomerations often have

    access to dense local networks, including frequent businessmeetings with other companies located in the same cluster.This is dierent for rms located in small agglomerations.With the Internets ability to cross geographical distance bet-ter than previous communication technology (e.g.,Cairncross, 1997; Forman & Van Zeebroeck, 2012;Friedman, 2005), rms located in smaller agglomerations con-sequently stand to benet more. 4

    Third, smaller-sized rms have fewer employees, lower rev-enues, and often smaller R&D investments in absolute terms.Therefore, agglomeration benets from their own R&Dinvestments are lower as is the availability of internal sourcesof knowledge (Cohen, 2010; Klepper & Simons, 2005). Smallerrms may consequently benet more from Internet-enabledknowledge spillovers than larger rms (cf. Acs, Audretsch, &Feldman, 1994). Single-plant establishment may also havemore to gain from their industrys adoption of the Internetthan multi-plant establishments.Fourth, informal businesses may also have larger benets

    from Internet-enabled knowledge spillovers. These rms havefewer resources to build knowledge networks and are oftenexcluded from formal businesses networks. There is evidenceto show ICT benet informal businesses (e.g., Jensen, 2007;Muto & Yamano, 2009 and references provided in the intro-duction).

    (e) Knowledge spillovers and absorptive capacities

    Firms need the capacity to apply the knowledge they gainaccess to. If absorptive capacities are weak, knowledge spil-lovers eects are much lower or absent (cf. Gorg &Greenaway, 2004). Indigenous capacities are needed becauseinnovations developed elsewhere are often inappropriate inspecic rm contexts, unless incremental innovations to adjustthem are undertaken (Atkinson & Stiglitz, 1969). The empiri-cal evidence conrms rms own capacities complements accessto knowledge (Hu, Jeerson, & Jinchang, 2005; Kokko, 1994;Kokko, Tansini & Zejan, 1996). Thus, industries adoption ofthe Internet may have heterogeneous eects on rms at dier-ent productivity levels: the most productive rms with stron-ger absorptive capacities may benet more from knowledgespillovers than less productive rms that lack such capacities.

    (f) Testable hypotheses

    Based on the discussion, the following testable hypothesesfor our empirical analysis arise:

    1. We expect the use of the Internet by industries as a toolfor communication to have positive impacts on rmsproductivity and innovation performance.

    2. We hypothesize that the Internet has heterogeneousimpacts on groups of rms that engage to dierentextents in innovation: (i) exporting and non-exportingrms, (ii) rms located in larger and smaller agglomera-tions, (iii) single- and multi-product rms and (iv) dier-ently sized rms. We also analyze (v) whether informalbusinesses benets from their industries adoption ofthe Internet.

    3. We hypothesize that the impact of industries adoptionof the Internet on rms productivity and innovationperformance diers for rms of dierent productivity

    590 WORLD DElevels to proxy for rms dierent levels of absorptivecapacities.Total services 21,061 42.1

    Size

    Micro (110 employees) 16,549 33.1Small (1150 employees) 20,022 40.0Medium (51150 employees) 7,772 15.5Large (more than 150 employees) 5,670 11.3

    Year

    2006 12,280 24.62007 8,261 16.52008 2,382 4.82009 14,057 28.12010 11,182 22.42011 1,851 3.7

    Income level

    High income 2,627 5.3Upper-middle income 21,126 42.2Lower-middle income 17,925 35.8Low income 8,335 16.7Full sample 50,013

  • Interestingly for our purposes, the WBES has informationon rms use of the Internet whether rms used email tocommunicate with suppliers and customers rather than oftheir investment in ICT, which says little about actual use.The indicator informs about whether the Internet is used forcommunication, which is critical for our study since it relatesto rms access to knowledge.What is more, rms email use is positively correlated with

    the intensity of their use of ICT in business operations.Table 2 shows results using linear probability, logistic andprobit regression models for the sample of rms for whichwe have information on whether they use the Internet to con-duct R&D, purchase inputs from suppliers, deliver services toclients, and market the rm on a website. 7 Results reportedin columns (1), (2), and (3) of Table 2 indicate that rmemail use is positively correlated with advanced uses of theInternet i.e., those using the Internet in all these ways. Bycontrast, as shown in columns (7), (8), and (9) of Table 2,rm email use is negatively correlated with weak uses of

    equipment as well as their ownership of quality certicates andpatents. (Information on patenting is unfortunately only avail-

    Y ijct a b1 ICT jct b2 ICT ijct C X ijct kj kct eijct1

    where Yijct is a measure of rm is labor productivity or itsinnovation performance, i.e., whether the rm owns aquality certicate or patent and its level of equipmentinvestment. Coecient b1 is our variable of interest toidentify knowledge-spillover eects from the uptake of theInternet: ICTjct is an indicator of industry js use of emailto communicate with clients and suppliers in country c inyear t. This measure is built for industries (by country-year) with at least 10 observations. ICTijct is a measureof rm is use of the Internet to control for direct impactsof rms Internet adoption while kj and kct are respectivelya set of industry and country-year dummies. In conse-quence, our identication strategy exploits dierences inindustrys adoption of the Internet across countries whilecontrolling for characteristics specic to industries andcountries in any year.We obtain a measure of country-year industry adoption,

    tens

    PM(4)

    65*

    .024eseses

    ,678.04

    hete. T(2)

    HAS THE INTERNET FOSTERED INCLUSIVE INNOVATION IN THE DEVELOPING WORLD? 591able for a smaller number of observations because the variableis not collected across all WBES surveys we combine in ouranalysis.) The WBES also has ample information on rm char-acteristics, including on sales, employment, ownership type,and export performance.We also use the informal rm dataset, provided by the

    WBES, which covers 1,557 rms for seven countries 8 in2010. The informal business survey collects information onrms uptake of mobile phones as well as on their sales andtheir equipment investments.

    4. EMPIRICAL FRAMEWORK

    To study the impact of industries adoption of the Interneton rms innovation and productivity performance, we usethe following estimation model:

    Table 2. Correlations between dierent in

    Advanced use of ICT

    LPM Logit Probit L(1) (2) (3)

    Email use 0.223*** 1.920*** 1.049*** 0.1[0.405] [0.364]

    (0.013) (0.199) (0.098) (0Firm-level controls Yes Yes Yes YIndustry xed eects Yes Yes Yes YCountry-year xed eects Yes Yes Yes Y

    Observations 13,678 13,655 13,655 13R2 0.10 0Pseudo-R2 0.08 0.08

    Note: Advanced, median, and low uses of ICT are dened depending on wsuppliers, (iii) deliver services to clients and (iv) market the rm on a websitcolumns (1), (4), and (7) and for logistic and probit regressions in columnsthe Internet i.e., those using the Internet for none of the pur-poses described above. For median users the correlation ispositive but less strong than for advanced users (columns(4), (5), and (6) of Table 2).The WBES also has information on rms labor productiv-

    ity and on innovation activities including their investments inshown in parentheses. For logistic and probit regressions, marginal eects are reof Panel A of Table 4. ***, **, and * indicate signicance at 1%, 5%, and 10%identied across 15 dierent industries. Xijct is a vector ofrm-level control variables as described in Section 6. Acset al. (1994) and Haskel, Pereira, and Slaughter (2007) use sim-ilar empirical methodologies to study respectively the impactsof industries R&D and FDI intensities. 9

    Three possible challenges may be raised with regard to theproposed empirical model: (i) endogeneity, that is, while ICTmight support innovation performance, more innovative rmsalso rely more on ICT, (ii) omitted variable bias, i.e., theremight be other factors that aect the productivity-ICT adop-tion relationship and (iii) measurement error of the explana-tory variable.First, endogeneity is less of a challenge for our analysis

    compared to an analysis of rms own investment in ICTon their performance. It is unlikely that rms innovationand productivity performance has a direct impact on theirindustrys adoption of the Internet. To avoid potentialendogeneity concerns, we exclude rm is own use of theInternet from the industry average we compute. Notwith-standing, we cannot exclude a small risk of endogeneity,which could arise if the most productive and innovativerms adoption of the Internet led other rms to usethe Internet. Higher industry adoption rates could thenbe correlated with strong performance and explain possible

    ities of rms ICT use and their email use

    Dependent variables:

    Median use of ICT Low use of ICT

    Logit Probit LPM Logit Probit(5) (6) (7) (8) (9)

    ** 0.690*** 0.426*** 0.388*** 1.918*** 1.147***[0.166] [0.165] [0.195] [0.216]

    ) (0.104) (0.064) (0.023) (0.108) (0.064)Yes Yes Yes Yes YesYes Yes Yes Yes YesYes Yes Yes Yes Yes

    13,677 13,678 13,677 13,6770.11

    0.08 0.08 0.12 0.12

    her rms used the Internet to: (i) conduct R&D, (ii) purchase inputs fromhe table reports results from linear probability model regressions (LPM) in, (5), and (8) and (3), (6), and (9), respectively. Robust standard errors are

    ported in brackets. Firm-level controls are the same as those of column (5)condence levels, respectively.

  • positive coecients on industry adoption rates. Weaddress endogeneity concerns as part of our robustnesstests.Second, we address omitted variable biases by introducing

    industry and country-year xed eects in addition to rm-level controls, described in Appendix Table 11A. Country-year xed eects allow isolating potential dierences acrosscountries in specic years. This includes government policieswith possible impacts on rms productivity and innovationperformance. Controlling for industry eects is also importantbecause certain industries are more technology-intensive thanothers.Third, measurement error is less of a concern than for

    analyses that focus on interpreting rm-level evidence. Ourvariable of interest is aggregated at the industry level andconsequently possible misreporting at rm-level is better

    includes all explanatory variables as in (1), (2), and (3)(Koenker & Bassett, 1978). Estimating h from 0 to 1 givesthe entire conditional distribution of Prodict, conditional onzijct (Buchinsky, 1998). In other words, using quantileregressions shows the impact of industries Internetadoption at dierent levels of the conditional productivitydistribution, rather than at the conditional mean of thedependent variable. Empirical applications of quantileregression techniques include, for example, Yasar andMorrison Paul (2007), Fattouh, Scaramozzino, and Harris(2005) and Coad and Rao (2008). Quantile regressions alsoallow for a robustness test of our main results as they are lesssensitive to outliers than standard regression models (Koenker& Bassett, 1978). 10

    log

    n

    9543

    6973

    7591

    592 WORLD DEVELOPMENTcontrolled for.In order to test for possible heterogeneous eects across

    rms we estimate the following modied model:

    Y ijct a bType1 ICT jct Typeijct b1 ICT jct b2 ICT ijct C Xijct kj kct eijct 2

    Y ijct a bADV 1 ICT jct TypeADV ijct bDIS1 ICT jct TypeDISijct b2 ICT ijct C Xijct kj kct eijct 3

    where Typeijct indicates rm characteristics (such as rm issize) and TypeADVijct and TypeDISijct are dichotomous vari-ables of rm characteristics (for instance, whether the rm isan exporter, TypeADVijct, or not, TypeDISijct).We use various estimation models in our analysis. To esti-

    mate impacts on the average rms labor productivity andequipment investment we apply ordinary least squares regres-sions. We use logistic and probit estimation models to assessthe impacts of industry Internet adoption on the two binaryoutcome variables: quality certicates and patents. Robuststandard errors clustered at country-industry-year level areapplied systematically to account for the fact that industryadoption of the Internet is an aggregate variable (Moulton,1990).Moreover, we apply quantile regressions to assess whether

    impacts dier based on rms level of labor productivity.Quantile regressions can be expressed in the general formProdict = x

    0ictb + eict with Qh (Prodict/zijct) = z

    0ijctbh, where zijct

    Table 3. Statistics on techno

    Overall

    Firm nbr. % Firm

    Use of cell-phone

    No 1,026 40.7 2Yes 1,495 59.3 9

    Use of electricity

    No 553 24.9 3Yes 1,668 75.1 8

    Experienced power outages

    No 765 46.1 2Yes 894 53.9 5Note: The statistics are based on rm observations for 14 countries: Angola, ARepublic of Congo, Ivory Coast, Guatemala, Madagascar, Mali, Mauritius, Ny use of the informal sector

    AFR LAC

    br. % Firm nbr. %

    23.8 674 58.076.2 489 42.1

    29.7 178 20.770.3 681 79.3

    31.8 489 72.068.2 190 28.05. DESCRIPTIVE STATISTICS

    Many rms in developing and emerging countries haveadopted the Internet to support their operations over the200611 period, but uptake rates diered between coun-tries and rms. While 47% and 57% of the rms inlow-income and lower-middle-income economies communi-cated with clients and supplier by email, 84% and 93%of the rms in upper-middle-income and high-incomeeconomies did so. Moreover, small rms were less activeusers than larger businesses: their uptake was of 44.5%compared to 96.9% for businesses with more than 150employees. Informal businesses were also active users ofmobile phones. Table 3 shows that 76.2% of the Africanbusinesses in our sample used mobile phones in 200910even though more than two thirds of these rms experi-enced power outages and more than one in four rmsdid not have electricity.The use of the Internet varied across dierent countries

    industries. In the textiles industry, for instance, the share ofrms using the Internet for communication was of only21% in Nigeria, 25% in Indonesia, and 33% in Pakistan whilethe share was of 100% in Argentina, Costa Rica, and Peru.In the retail and wholesale trade sector, the same sharesrange from 20% for Uzbekistan and 30% for Angola toalmost complete adoption in Hungary (96%) and Estonia(99%). Figure 1 shows substantial dispersion existed acrosscountries in industries adoption rates. 11 In the food, gar-ment, and service industries i.e., retail and wholesale tradeand also hotels and restaurants several industries hadweakly adopted the Internet. By contrast, chemicals andrgentina, Botswana, Burkina Faso, Cameroon, Cape Verde, Democraticepal, and Peru.

  • rms employment and age (column 2), indicators of public

    INownership and whether the establishments are part of multi-plant establishments (column 3) and controls for whether therm has connections abroad (i.e., foreign-ownership andpharmaceuticals industries had high adoption rates in mostcountries.

    6. RESULTS

    (a) Baseline results: ICT-enabled spillovers on rm productivityand innovation performance

    First, we test whether the wider diusion of ICT leads toknowledge spillovers that result in higher rm productivityand improved innovation performance. Panel A of Table 4shows regression results of Eqn. (1) for labor productivity: col-umn (1) reports results for industries use of the Internet withindustry and country-year xed eects. We identify a positivesignicant eect. We progressively add rm-level controls:

    0

    20

    40

    60

    80

    100 Max75th percentileMedian

    25th percentile

    Min

    Perc

    enta

    geof

    firm

    s usi

    ng th

    e In

    tern

    et

    Figure 1. Dierences in industries adoption of the Internet across countries.Note: The deciles for dierent industries are computed based on thefollowing number of country observations on the share of rms using theInternet to communicate by email with clients and users: 110 for food, 50for textiles, 71 for garments, 48 for chemicals and pharmaceuticals, 68 formetals and machinery, 64 for non-metallic and plastic materials, 123for retail and wholesale trade, 74 for hotels and restaurants and 81 forconstruction and transportation. Statistics provided are obtained for the

    50,013 rms included in our baseline sample.

    HAS THE INTERNET FOSTERED INCLUSIVEexporter status) (column 4). We also add proxies for manage-rial quality and access to nance (column 5). Consistently withthe literature, we nd that these factors are positively corre-lated with rms productivity. Only public ownership is nega-tively correlated with rms productivity. We also control forrms own use of the Internet to exclude direct eects on rms(column 6). Our variable of interest, the industry-wide adop-tion of the Internet as a means of communication, remainspositive signicant but decreases as other factors, notablyrms own use of the Internet, are added to the specication.Appendix Table 11B provides a correlation matrix of the inde-pendent variables; levels of correlation are modest and justifythe use of all dependent variables in our empirical estima-tions. 12

    Panel B of Table 4 shows positive signicant eects on aver-age rms investment in equipment (column 1). We also iden-tify positive eects on rms ownership of quality certicates(columns 2 and 3) and patents (columns 4 and 5). Both resultsfrom probit and logistic estimation models are included. Forbrevity we only report results from logistic estimation modelsin subsequent tables. 13

    Overall, our results provide evidence that industries adop-tion of the Internet facilitates positive spillover eects onrms productivity and innovation performance. Resultsconrm our rst empirical hypothesis. As for the magnitudeof estimated eects, all else equal, an increase by one standarddeviation in the intensity of a industries use of the Internetimproves rm labor productivity by an amount equivalentto productivity increasing from the 50th to the 54th percentileof the distribution. For equipment investment the increase isfrom the 50th to the 55th percentile of the corresponding distri-bution. Impacts on rms ownership of quality certicates andpatents are modest. An increase by one standard deviationwould, all else equal, lead to an increase in formal intellectualproperty rights ownership of 3% and 5% respectively.

    (b) Robustness tests

    This section presents robustness tests of our results. Find-ings for labor productivity are reported in Table 5. First, inorder to ensure that our variable of interest does not pickup the eects of other industry characteristics, we include mea-sures of national industries average employment, their age,foreign ownership status, the volume of exporter activities,an indicator of public ownership, the share of multi-plantestablishments, the average of years of managers experience,and an indicator of credit access. Results, reported in column(1) of Table 5, conrm our evidence is robust to includingthese control variables. Unreported tests show our results alsohold if we include location xed eects and rms past produc-tivity performance to account for a variety of possible omittedfactors.Second, we check whether our results are robust to potential

    outliers. We exclude the 5% most and least productive rmsfor each country-year group. 14 Results, reported in column(2) of Table 5, show outliers do not drive results. Results fromquantile regressions, reported in Section 6(d), provide addi-tional evidence that outliers do not aect our ndings. Arelated concern arises in those cases where few observationsare used to obtain averages of industry Internet adoption.We, therefore, raise the threshold for averages to 30 observa-tions. Results shown in column (3) of Table 5 conrm ourresults are robust. The downside to raising the threshold isthat fewer observations can be included as part of the analysis.This introduces a potential sample selection bias. This is whywe allow for a lower threshold for our main results. Third, ourmain results focus on spillover eects within the rms indus-try, as product markets are sources of relevant knowledge forrms productivity and innovation performance. The mostadequate sources of knowledge for rms may, however, benot be rms own industries. In order to account for possiblealternative sources for knowledge spillovers we compute alter-native measures, including separate measures of industriesadoption of the Internet for smaller and large rms. Smallerrms may have more to gain from other rms of similar sizeas processes adopted by large rms may be out of reach forthem. By contrast, large rms may have little to gain fromknowing about the practices adopted by smaller entities.Results, reported in column (4) of Table 5, are positive signif-icant and conrm our main ndings.Moreover, we obtain separate measures of industries adop-

    tion of the Internet for dierent types of locations as inFisman and Svensson (2007). Firms in rural areas with veryfew inhabitants may have more to gain from the practices ofother rms located in similar types of locations. 15 We nd apositive signicant impact (column (5) of Table 5). In addi-tion, unreported results using a measure of Internet adoptionfor country-location-year level (as in Arnold, Mattoo, &

    NOVATION IN THE DEVELOPING WORLD? 593Narciso, 2008, and Dollar et al., 2006), are also positive signif-icant.

  • ase

    Dep

    VETable 4. B

    Panel A: Labor productivity

    (1) (2)

    Industry Internet use 0.010*** 0.008***

    (0.001) (0.001)

    Firm-level controls

    Employment 0.151***

    (0.010)***

    594 WORLD DEFinally, we compute a measure of Internet adoption at thecountry level. This specication does not allow includingcountry xed eects. In order to ensure that other dierencesin countries business environment do not aect results, weadd a comprehensive set of country controls to our specica-tions: GDP per capita, gross capital formation, net foreigndirect and portfolio investment, domestic credit to the privatesector (as % of GDP), school enrollment in primary and sec-ondary education, literacy rates as well as world region dum-mies. Our results reported in column (6) of Table 5 are positivesignicant.Fourth, we test whether exposure to better technology

    improves impacts on rm productivity and innovation

    Age 0.083(0.011)

    Public ownership

    Multi-plant rm

    Foreign ownership

    Exporter status

    Credit access

    Managerial expertise

    Firm-level Internet use

    Internet use

    Industry xed eects Yes YesCountry-year xed eects Yes Yes

    Observations 56,169 55,121R2 0.79 0.80Panel B: Innovation performance

    Equipment investment

    OLS(1)

    Industry Internet use 0.009***

    (0.003)Firm-level controls YesFirm-level Internet use YesIndustry xed eects YesCountry-year xed eects Yes

    Observations 33,080R2 0.45Pseudo R2

    Note: Panel A reports results from ordinary least squares regressions. Robuparentheses. For logistic and probit regressions, marginal eects are reported i(6) of Panel A. ***, **, and * indicate signicance at 1%, 5%, and 10% condline results

    endent variable: Labor productivity

    (3) (4) (5) (6)

    0.008*** 0.007*** 0.007*** 0.006***

    (0.001) (0.001) (0.001) (0.001)

    0.132*** 0.082*** 0.058*** 0.023**

    (0.010) (0.010) (0.010) (0.010)*** *** *** ***

    LOPMENTperformance. We interact our variable of interest with ameasure of industries use of imported technologies. As shownin column (7) of Table 5, we nd that spillover returns fromthe Internet are larger where the exposure to technology ismore important.Finally, we check whether our results are dierent for rms

    in the manufacturing and services sectors. As shown in column(8) of Table 5, we nd positive signicant eects for both typesof rms but larger returns for services rms. This may bebecause knowledge is more important for services than formanufacturing rms. Alternatively, services rms may benetmore because better knowledge about customer preferencesmay be more critical for them than for manufacturing rms.

    0.078 0.087 0.078 0.075(0.011) (0.011) (0.011) (0.011)0.133* 0.149** 0.121* 0.136*(0.073) (0.072) (0.073) (0.071)0.333*** 0.280*** 0.285*** 0.254***

    (0.022) (0.022) (0.022) (0.022)0.443*** 0.476*** 0.453***

    (0.025) (0.026) (0.025)0.258*** 0.241*** 0.191***

    (0.022) (0.022) (0.021)0.302*** 0.279***

    (0.016) (0.016)0.026** 0.027**

    (0.011) (0.011)

    0.386***

    (0.017)Yes Yes Yes YesYes Yes Yes Yes

    52,839 52,146 50,107 50,0130.80 0.81 0.81 0.81

    Quality certicates Patents

    Probit Logistic Probit Logistic(2) (3) (4) (5)

    0.003** 0.005** 0.007** 0.012**

    [0.001] [0.002] [0.002](0.001) (0.002) (0.003) (0.005)Yes Yes Yes YesYes Yes Yes YesYes Yes Yes YesYes Yes Yes Yes

    54,586 54,586 9,061 9,061

    0.25 0.25 0.19 0.19

    st standard errors clustered at country-industry-year level are shown inn brackets. Firm-level controls in Panel B are the same as those of columnence levels, respectively.

  • Table 5. Robustness tests

    Dependent variable: Labor productivity

    Controls Controlling for outliers Alternative aggregation Extensions

    Addingcontextcontrols

    Removing thetop and bottom

    5%

    Setting thethreshold atN P 30

    Firm size Locationtype

    Countrylevel

    Exposureto

    technology

    Manufacturingvs. services

    (1) (2) (3) (4) (5) (6) (7) (8)

    Industry Internet use 0.005*** 0.005*** 0.006***

    (0.001) (0.001) (0.002)Industry Internet use (rm size) 0.007***

    (0.001)Industry Internet use (location type) 0.009***

    (0.001)Country Internet use 0.244***

    (0.018)Industry Internet use * high exposure to technology 0.006***

    (0.001)Industry Internet use * low exposure to technology 0.005***

    (0.001)Industry Internet Use * manufacturing 0.005***

    (0.001)Industry Internet Use * services 0.008***

    (0.001)P-value for the dierence in coecients 0.00 0.01Firm-level controls Yes Yes Yes Yes Yes Yes Yes YesFirm-level Internet use Yes Yes Yes Yes Yes Yes Yes YesIndustry xed eects Yes Yes Yes Yes Yes Yes Yes YesCountry-year xed eects Yes Yes Yes Yes Yes No Yes YesCountry-level controls No No No No No Yes No No

    Observations 49,790 44,959 44,115 44,476 41,442 9,777 50,013 50,013R2 0.81 0.87 0.81 0.82 0.82 0.83 0.81 0.81

    Note: The table reports results from ordinary least squares regressions. Firm-level controls are the same as those of column (6) of Panel A of Table 4. Country controls included in regressions reported incolumn (6) are country GDP per capita, the level of gross capital formation, national foreign direct and portfolio investment, domestic credit to private sector (as % of GDP), country school enrollmentin primary and secondary education as well as country literacy rates. Robust standard errors clustered at level of the Internet use variable are shown in parentheses. ***, **, and * indicate signicance at1%, 5%, and 10% condence levels, respectively.

    HASTHEIN

    TERNETFOSTERED

    INCLUSIVEIN

    NOVATIO

    NIN

    THEDEVELOPIN

    GWORLD?

    595

  • Robustness tests for our measures of innovation perfor-mance are reported in Appendix Table 12. As shown in PanelA, robustness tests conrm ndings regarding rms invest-ments in equipment. In this case, we do not nd impacts tobe dierent with regard to the exposure to technology (column7), or across manufacturing and services rms (column 8).With regard to quality certicates and patents, Panels B andC of Appendix Table 12 show the evidence is less robust thanthat on productivity and equipment investments. In the case ofquality certicates, manufacturing rms benet more than ser-vices rms. 16

    (c) Testing for heterogeneous impacts of the Internet acrossrms

    We test our second hypothesis on dierential gains fromindustry Internet adoption across dierent groups of rms.We test for dierences in impacts across (i) exporters andnon-exporters, (ii) rms located in larger and smalleragglomerations, (iii) single- and multi-product rms, and(iv) smaller and larger rms. We also analyze whether (v)informal businesses benet from their industries adoptionof the Internet.Table 6A reports results for impacts on labor productivity.

    We nd that there is more to be gained for non-exporters(column 1 of Table 6A). Unreported results indicate thatnational rms also benet more than foreign-owned rms.Column (2) of Table 6A shows results where we distinguishthose rms located in countries capitals or in cities of morethan 1 million inhabitants from those located in smalleragglomerations. We nd statistically signicant stronger

    impacts on labor productivity for rms in small agglomera-tions, even if location eects are controlled for. Column (3)of Table 6A shows that single-plant rms benet more com-pared to multi-plant rms, but the dierence is not statisti-cally signicant. Finally, with regard to dierently sizedrms, reported in column (4) of Table 6A, we do not ndevidence of heterogeneous eects when it comes to labor pro-ductivity.The evidence for innovation indicators also points to dif-

    ferential eects, but is more mixed. We identify no eectson exporters, as shown in columns (1), (5), and (9) ofTable 6B. The dierence in coecients is statistically signif-icant for equipment investment and patenting. As forlocation, while we nd that rms in small agglomerationshave larger returns to innovation performance, thedierence in coecients is statistically signicant for rmsownership of quality certicates. Results are shown incolumns (2), (6), and (10) of Table 6B. As reported incolumns (3), (7), and (11) of Table 6B we also ndsingle-plant rms are more likely to obtain quality certi-cates and patents as a result of their industries adoptionof the Internet. We do not identify dierential impacts onequipment investment rates. As for rm size dierences,results reported in columns (4), (8), and (12) show nosignicant dierences in impacts except for results on rmquality certicate adoption.Finally, column (1) of Table 7 shows that informal busi-

    nesses also benet from ICT-enabled knowledge spillovers interms of sales gains. We use the industrys adoption of mobilephones as a proxy for the stronger use of ICT by informalbusinesses. The evidence is maintained if we control for

    rom

    irm

    596 WORLD DEVELOPMENTTable 6A. Firm characteristics and benets f

    Exporters(1)

    Industry Internet use * exporters 0.003*

    (0.002)Industry Internet use * non-exporters 0.006***

    (0.001)Industry Internet use * big agglomeration

    Industry Internet use * small agglomeration

    Industry Internet use * multi-plant rms

    Industry Internet use * single-plant rms

    Industry Internet use * bigger rms

    Industry Internet use * smaller rms

    Firm-level controls YesFirm-level Internet use YesIndustry xed eects YesCountry-year xed eects YesP-value of the dierence in coecients 0.00

    Observations 50,013R2 0.81

    Note: The tables reports results from ordinary least squares regressions. F

    Robust standard errors clustered at country-industry-year level are showncondence levels, respectively.the Internets adoption: Labor productivity

    Dependent variable: Labor productivity

    Firm location Multi-plant rms Firm size(2) (3) (4)

    0.005***

    (0.001)0.008***

    (0.001)0.004***

    (0.002)0.006***

    (0.001)0.006***

    (0.001)0.006***

    (0.001)Yes Yes YesYes Yes YesYes Yes YesYes Yes Yes0.01 0.08 0.45

    44,706 50,013 50,0130.82 0.81 0.81

    -level controls are the same as those of column (6) of Panel A of Table 4.

    in parentheses.***, **, and * indicate signicance at 1%, 5%, and 10%

  • Table 6B. Firm characteristics and benets from the Internets adoption: Innovation performance

    Dependent variables

    Equipment investment Quality certicates Patents

    OLS regressions Logistic regressionsExporters Firm

    locationMulti-plant rms Firm size Exporters Firm

    locationMulti-plant rms Firm size Exporters Firm

    locationMulti-plant rms Firm size

    (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

    Industry Internetuse * exporters

    0.001 0.003 0.002

    [0.000] [0.000](0.005) (0.003) (0.006)

    Industry Internetuse * non-exporters

    0.010*** 0.006*** 0.015***

    [0.001] [0.003](0.003) (0.002) (0.005)

    Industry Internetuse * big agglomeration

    0.008** 0.002 0.010**

    [0.000] [0.002](0.004) (0.002) (0.005)

    Industry Internetuse * small agglomeration

    0.011** 0.006** 0.014**

    [0.001] [0.003](0.004) (0.003) (0.006)

    Industry Internetuse * multi-plant rms

    0.009*** 0.002 0.005

    [0.000] [0.001](0.004) (0.002) (0.006)

    Industry Internetuse * single-plant rms

    0.009*** 0.007*** 0.013**

    [0.001] [0.002](0.003) (0.002) (0.005)

    Industry Internetuse * bigger rms

    0.009*** 0.006*** 0.011**

    [0.001] [0.002](0.003) (0.002) (0.005)

    Industry Internetuse * smaller rms

    0.009*** 0.004** 0.013**

    [0.001] [0.002](0.003) (0.002) (0.005)

    Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesFirm-level Internet use Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesIndustry xed eects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesCountry-year xed eects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes YesP-value of the dierence in coecients 0.01 0.44 0.89 0.93 0.18 0.08 0.00 0.01 0.00 0.31 0.05 0.31

    Observations 31,281 27,612 31,281 31,281 54,586 49,048 54,586 54,586 9,061 9,061 9,061 9,061R2 0.45 0.44 0.45 0.45Pseudo R2 0.25 0.25 0.25 0.25 0.19 0.19 0.19 0.19

    Note: Firm-level controls are the same as those of column (6) of Panel A of Table 4. Robust standard errors clustered at country-industry-year level are shown in parentheses. For logistic regressions,marginal eects are reported in brackets. ***, **, and * indicate signicance at 1%, 5%, and 10% condence levels, respectively.

    HASTHEIN

    TERNETFOSTERED

    INCLUSIVEIN

    NOVATIO

    NIN

    THEDEVELOPIN

    GWORLD?

    597

  • rm-level control variables employment size, age, ownershipof bank accounts, and whether rms had a loan (column (2)).

    in the productivity gains from industries adoption of theInternet. Gains are small for rms with productivity levels

    Table 7. The impact of industry cell phone use on the performance of informal businesses

    Dependent variables

    Sales Machinery investment

    (1) (2) (3) (4) (5) (6)

    Industry cell phone use 0.010*** 0.011*** 0.010*** 0.017** 0.016* 0.013(0.003) (0.003) (0.003) (0.009) (0.009) (0.009)

    Firm-level controls No Yes Yes No Yes YesFirm-level cell phone use No No Yes No No YesIndustry xed eects Yes Yes Yes Yes Yes YesCountry-year xed eects Yes Yes Yes Yes Yes Yes

    Observations 1,406 1,207 1,207 1,430 1,219 1,219R2 0.80 0.83 0.84 0.09 0.14 0.15

    Note: The table reports results from ordinary least squares regressions. Firm-level controls include employment size, their age, their ownership of bankaccounts and whether they had a loan. Robust standard errors are shown in parentheses. ***, **, and * indicate signicance at 1%, 5%, and 10%condence levels, respectively.

    598 WORLD DEVELOPMENTResults are also robust to controlling for rms use of cellphones (column (3). 17 However, the evidence of positiveimpacts on informal rms machinery investments is weak(columns (4)(6) of Table 7). Our estimated eects becomeinsignicant once we control for rms own use of cell phones(column (6) of Table 7).In conclusion, with regard to our second hypothesis, we nd

    that the Internet provides larger opportunities for productivityimprovements of groups of rms that have limited access tooine knowledge networks. This, however, does not holdfor dierently sized rms.

    (d) Testing for the eects of absorptive capacities

    We test our third hypothesis on the role of rms absorp-tive capacities for positive impacts from industry Internetuse on rm productivity and innovation performance. Inorder to test whether the average rm eects identied hidedierences in benets across rms of dierent productivitylevels, we conduct quantile regressions of Eqn. (1). Results,which are reported in Figure 2, show that there are dierences0.002

    0.003

    0.004

    0.005

    0.006

    0.007

    0.008

    0.009

    0.01

    10 15 20 25 30 35 40 45

    Coe

    fficient

    Qu

    Figure 2. Impacts of the Internets adoption on productivity across the rm proNote: The gure plots coecients from quantile regressions of the impact of th

    quantile of the producbelow the 35th percentile but increase for more productiverms, before leveling o at the 70th percentile. This evidencesupports our hypothesis that rms absorptive capacitiesmatter; the least productive rms have limited returns fromtheir industries Internet adoption.In addition, we test whether average impacts across the rm

    characteristics reported in Section 6(c) dier across the pro-ductivity distribution. With regard to exporter status, Figure 3(a) shows highest dierential returns for non-exporting rmsat productivity levels above the median. While the gains fornon-exporters rise across the productivity distribution, bene-ts for exporters are low for all producers, including the mostproductive rms. In other words, the larger average impactson non-exporters performance identied in our previous anal-ysis are driven by larger productivity gains for the most pro-ductive non-exporters. There is little dierence in (low)benets among the least productive exporters and non-exporters.Figure 3(b) plots the coecients of our variable of interest

    for rms located in dierent agglomerations. The least produc-tive rms reap minor productivity gains from industries Inter-50 55 60 65 70 75 80 85 90antiles

    ductivity distribution.

    e share of rms using email on labor productivity from the 10th to the 90th

    tivity distribution.

  • (a) By exporter status

    (b) By agglomeration type

    (c) By single- and multi-plant firms

    0.001

    0.002

    0.003

    0.004

    0.005

    0.006

    0.007

    0.008

    0.009

    0.010

    10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

    Coe

    fficient

    Quantiles

    Non-exporters

    Exporters

    0.000

    0.002

    0.004

    0.006

    0.008

    0.010

    0.012

    10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

    Coe

    fficient

    Quantiles

    Firms in small agglomerations

    Firms in big agglomerations

    0.001

    0.002

    0.003

    0.004

    0.005

    0.006

    0.007

    0.008

    0.009

    10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

    Coe

    fficient

    Quantiles

    Single-plant firms

    Multi-plant firms

    Figure 3. Dierent types of rms and the impact of the Internets adoption on productivity across the rm productivity distribution.Note: The gures plots coecients from quantile regressions of the impact of industry in big and small agglomerations as in column (3) of Panel A of

    Table 5 on labor productivity from the 10th to the 90th quantile of the distribution.

    HAS THE INTERNET FOSTERED INCLUSIVE INNOVATION IN THE DEVELOPING WORLD? 599

  • net adoption. By contrast, returns are larger for rms withproductivity levels above the median with larger gains forrms located in smaller agglomerations. The gap betweenrms located in bigger and smaller locations is largest for rmsin the median productivity range. Firms at higher levels of

    productivity do not benet more from knowledge spilloversthan those at median productivity.Figure 3(c) reports results for multi- and single-plant rms.

    The evidence shows that the Internet provides limited returnsto the least productive single- and multi-plant rms. However,

    Table 8. The eect of size on the impact of the Internets adoption across the productivity distribution

    Dependent variable: Labor productivityQuantile regression

    Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9(1) (2) (3) (4) (5) (6) (7) (8) (9)

    Industry Internet use * size 0.000 0.001** 0.002*** 0.002*** 0.002*** 0.003*** 0.003*** 0.003*** 0.003***(0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)

    Industry Internet use 0.003 0.007*** 0.009*** 0.011*** 0.012*** 0.014*** 0.015*** 0.015*** 0.016***

    (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) (0.003)Firm-level controls Yes Yes Yes Yes Yes Yes Yes Yes YesFirm-level Internet use Yes Yes Yes Yes Yes Yes Yes Yes YesIndustry xed eects Yes Yes Yes Yes Yes Yes Yes Yes YesCountry-year xed eects Yes Yes Yes Yes Yes Yes Yes Yes Yes

    Observations 50,107 50,107 50,107 50,107 50,107 50,107 50,107 50,107 50,107R2 0.78 0.80 0.81 0.81 0.81 0.81 0.81 0.80 0.78

    Note: Firm-level controls are the same as those of column (6) of Panel A of Table 4. Robust standard errors clustered at country-sector-year level areshown in parentheses. ***, **, and * indicate signicance at 1%, 5%, and 10% condence levels, respectively.

    Table 9. Impacts of productivity dierences on the Internet adoptions impact on rm innovation performance

    Equipmentinvestment

    Quality certicates Patents

    OLS regressions Logistic regressions

    (1) (2) (3) (4) (5) (6)

    Industry Internet use * below median 0.005 0.004* 0.009*

    [0.000] [0.002](0.003) (0.002) (0.005)

    Industry Internet use * above median 0.007* 0.005** 0.010**

    [0.001] [0.002](0.003) (0.002) (0.005)

    Industry Internet use * Q1 0.006 0.004* 0.011**

    [0.001] [0.002](0.003) (0.002) (0.005)

    Industry Internet use * Q2 0.004 0.004 0.008[0.000] [0.001]

    (0.003) (0.002) (0.005)Industry Internet use * Q3 0.006* 0.005* 0.010*

    [0.001] [0.002](0.003) (0.002) (0.005)

    Industry Internet use * Q4 0.008** 0.006*** 0.009*

    [0.001] [0.002](0.004) (0.002) (0.005)

    Firm-level controls Yes Yes Yes Yes Yes YesIndustry xed eects Yes Yes Yes Yes Yes YesCountry-year xed eects Yes Yes Yes Yes Yes YesP-value of dierence in coecients (below and above median) 0.02 0.02 0.47P-value of dierence in coecients (between Q1 and Q4) 0.12 0.02 0.40

    ,642.46

    f Tin b

    600 WORLD DEVELOPMENTObservations 26R2 0Pseudo-R2

    Note: Firm-level controls are the same as those of column (6) of Panel A oshown in parentheses. For logistic regressions, marginal eects are reported

    levels, respectively.26,642 41,720 41,720 7,087 7,0870.46

    0.26 0.26 0.18 0.18

    able 4. Robust standard errors clustered at country-industry-year level arerackets. ***, **, and * indicate signicance at 1%, 5%, and 10% condence

  • for the group of single-plant rms the benets rise for rms impact of industries Internet use on rm performance.

    provide larger benets to rms located in smaller agglom-

    and consequently Internet-enabled knowledge spillovers

    aimed at building rms absorptive capacities matter if the

    Several questions for future research arise, including how

    assess impacts. Informal rm surveys are also important to

    TE

    review of the literature.

    4. Forman, Goldfarb, and Greenstein (2014) conclude from their

    observations diers across industries.

    HAS THE INTERNET FOSTERED INCLUSIVE INNOVATION IN THE DEVELOPING WORLD? 601analysis of Internet investment and patenting across US counties, thatthe Internet has the potential to weaken the longstanding importance ofthe geographic localization of innovative activity (p. 5).

    5. Dethier et al. (2011) provide a review of the WBES.

    6. The routines used by the authors to clean the original dataset areavailable upon request.

    7. Only information on whether rms owned websites is also availablefor the large sample of rms.

    8. These countries are Angola, Argentina, Botswana, the DemocraticRepublic of Congo, Guatemala, Mali, and Peru.

    9. Seker (2012) and Dollar, Hallward-Driemeier, and Mengistae (2006)apply this approach to analyze the impacts of business conditions on rmperformance.3. User involvement can also stimulate innovation if it leads to co-innovation with users (Bresnahan, Brownstone, & Flamm, 1996; VonHippel, 2005).12. Unreported robustness tests show that results also hold ifonly the variable of interest i.e., industrys use of the Internet andrms own adoption of the Internet are included as part of theanalysis.

    13. All unreported results are available from the authors uponrequest.

    14. Our results are also robust to excluding outliers within industry-country-year categories.

    15. Unreported results for a measure of Internet adoption by location-type, rm size, sector, country, and year are also positive signicant.Aterido, Hallward-Driemeier, and Pages (2007) apply this approach intheir analysis.

    16. We cannot report similar tests for patents, for which we have mainlyinformation on manufacturing rms.

    17. We select a dierent set of control variables due to the dierentnature of rms analyzed and the dierent variables contained in theinformal rm survey.2. Cardona, Kretschmer, and Strobel (2013) provide a comprehensive11. As described in the notes of Figure 1 the number of countrywith higher than median productivity. By contrast, formulti-plant rms the rise is modest. As is the case of resultsfor exporters and non-exporters, we nd that the higher aver-age gains reported in column (3) of Table 6A is driven by thelarge returns for the most productive single-plant rms.Table 8 shows quantile regression estimates, which include a

    variable that interacts industry Internet adoption with rmsize. Dierently from average impacts reported in column (4)of Table 6A, quantile regression results indicate that thereare dierences across smaller and larger rms: more produc-tive smaller rms benet more than larger rms.Finally, with respect to impacts on innovation, we nd,

    as shown in Table 9, that the returns from industriesuse of the Internet on equipment investment and ownershipof quality certicates are larger for rms with above-median productivity than for those with below-medianproductivity. We nd no dierence with regard to rmpatenting.

    7. CONCLUDING REMARKS

    Using 50,013 rm observations for 117 countries overthe 200611 period, we provide evidence of a positive

    NO

    1. Ding, Levin, Stephan, and Winkler (2010) nd the Internet facilitatedthe inclusion of women scientists and those working at non-eliteinstitutions in collaborative research. Agrawal and Goldfarb (2008)show that the adoption of Bitnet, an early version of the Internet,disproportionately beneted middle-tier universities collaboration withleading universities.10. We analyze dierential impacts on other innovation variables byinteracting our variable of interest, industries adoption of the Internet,with above or below median rm productivity at t 3 or, respectively, thequartile of the distribution of productivity at t 3 the rm was part of(Table 9).Sexplore the potential of the Internet to help these businesses.Questions related to innovation should be added as these rmsalso engage in diverse non-technological innovation activities(OECD, 2015).increasingly sophisticated uses of the Internet inuence poten-tial spillovers and returns to rm performance. Ensuring rmsurveys capture these uses is critical for research to betterInternet is to support rms productivity and innovationperformance.help increase the group of innovating rms. Positive eectsare the more so notable as they also arise where rms facenancial constraints, frequent power outages, skills short-ages, corruption, and cumbersome labor regulations(Paunov & Rollo, 2015). However, complementary policieserations, to single-plant establishments and to non-exporters. These rms commonly engage less in innovationThese gains, which do not depend on rms own ICTinvestments, justify public policies aimed at fostering indus-tries use of the Internet. We also nd that the Internet

  • EBuchinsky, M. (1998). The dynamics of changes in the female wage FDI: A threshold regression analysis.Oxford Bulletin of Economics and

    VEdistribution in the USA: A quantile regression approach. Journal of Statistics, 67(3), 281306.REFER

    Acs, Z. J., & Audretsch, D. B. (1990). Innovation and small rms. MITPress.

    Acs, Z. J., Audretsch, D. B., & Feldman, M. P. (1994). R&D spilloversand recipient rm size. The Review of Economics and Statistics, 76(2),336340.

    Agrawal, A., & Goldfarb, A. (2008). Restructuring research: Communi-cation costs and the democratization of university innovation. TheAmerican Economic Review, 98(4), 15781590.

    Aker, J. C., & Mbiti, I. M. (2010). Mobile phones and economicdevelopment in Africa. Journal of Economic Perspectives, 24, 207232.

    Almeida, R., & Fernandes, A. M. (2008). Openness and technologicalinnovations in developing countries: Evidence from rm-level surveys.The Journal of Development Studies, 44(5), 701727.

    Arnold, J. M., Mattoo, A., & Narciso, G. (2008). Services inputs and rmproductivity in Sub-Saharan Africa: Evidence from rm-level data.Journal of African Economies, 17, 578599.

    Arrow, K. J. (1962). Economic welfare and the allocation of resources forinvention. In R. Nelson (Ed.), The rate and direction of inventiveactivity (pp. 609626). Princeton University Press.

    Arthur, W. B. (2007). The structure of invention. Research Policy, 36(2),274287.

    Aterido, R., Hallward-Driemeier, M., & Pages, C. (2007). Investmentclimate and employment growth; The impact of access to nance,corruption and regulations across rms. IZA Working Paper No. 3138.

    Atkinson, A. B., & Stiglitz, J. E. (1969). A new view of technologicalchange. Economic Journal, 79(315), 573578.

    Audretsch, D., & Feldman, M. (1996). R&D spillovers and the geographyof innovation and production. The American Economic Review,630640.

    Audretsch, D., & Feldman, M. (2004). Knowledge spillovers and thegeography of innovation. Handbook of Regional and Urban Economics,4, 27132739.

    Bartel, A., Ichniowski, C., & Shaw, K. (2007). How does informationtechnology aect productivity? Plant-level comparisons of productinnovation, process improvement, and worker skills. The QuarterlyJournal of Economics, 122(4), 17211758.

    Beck, T., Demirguc-Kunt, A., & Maksimovic, V. (2008). Financingpatterns around the world: Are small rms dierent?. Journal ofFinancial Economics, 89(3), 467487.

    Black, S., & Lynch, L. (2001). How to compete: The impact of work-placepractices and information technology on productivity. The Review ofEconomics and Statistics, 83(3), 434445.

    Black, S. E., & Lynch, L. M. (2004). Whats driving the new economy?:The benets of workplace innovation. The Economic Journal, 114(493),97116.

    Bloom, N., Sadun, R., & Van Reenen, J. (2012). Americans do it better:US multinationals and the productivity miracle. The AmericanEconomic Review, 102(1), 167201, February.

    Bloom, N., Schankerman, M., & Van Reenen, J. (2013). Identifyingtechnology spillovers and product market rivalry. Econometrica, 81(4),13471393.

    Bresnahan, T. F., Brynjolfsson, E., & Hitt, L. M. (2002). Informationtechnology, workplace organization, and the demand for skilled labor:Firm-level evidence. The Quarterly Journal of Economics, 117, 339376.

    Bresnahan, T., Greenstein, S., Brownstone, D., & Flamm, K. (1996).Technical progress and co-invention in computing and in the uses ofcomputers. Brookings Papers on Economic Activity. Microeconomics,183.

    Brynjolfsson, E., Hitt, L. M., & Kim, H. (2011). Strength in numbers: howdoes data-driven decision-making aect rm performance? MIT, unpub-lished manuscript.

    Brynjolfsson, E., & Yang, S. (1996). Information technology andproductivity: A review of the literature. Advances in Computers, 43,179214.

    602 WORLD DEApplied Econometrics, 13(1), 130.Cairncross, F. (1997). The death of distance. Harvard University Press.Cardona, M., Kretschmer, T., & Strobel, T. (2013). ICT and productivity:

    Conclusions from the empirical literature. Information Economics andPolicy, 25(2013), 109125.NCES

    Coad, A., & Rao, R. (2008). Innovation and rm growth in high-techsectors: A quantile regression approach. Research Policy, 37(4),633648.

    Coe, D. T., & Helpman, E. (1995). International R&D spillovers.European Economic Review, 39(5), 859887.

    Cohen, W. M. (2010). Fifty years of empirical studies of innovativeactivity and performance. Handbook of the Economics of Innovation, 1,129213.

    Cohen, W., & Levinthal, D. (1989). Innovation and learning: The twofaces of R&D. Economic Journal, 99, 569596.

    Collard-Wexler, A., Asker, J., & De Loecker, J. (2011). Productivityvolatility and the misallocation of resources in developing economies.National Bureau of Economic Research.

    Commander, S., Harrison, R., & Menezes-Filho, N. (2011). ICT andproductivity in developing countries: New rm-level evidence fromBrazil and India. The Review of Economics and Statistics, 93, 528541.

    Conley, T. G., & Udry, C. R. (2010). Learning about a new technology:Pineapple in Ghana. The American Economic Review, 3569.

    Dethier, J.-J., Hirn, M., & Straub, S. (2011). Explaining enterpriseperformance in developing countries with business climate survey data.World Bank Research Observer, 26, 258309.

    Ding, W. W., Levin, S. G., Stephan, P. E., & Winkler, A. E. (2010). Theimpact of information technology on academic scientists productivityand collaboration patterns. Management Science, 56(9), 14391461.

    Dollar, D., Hallward-Driemeier, M., & Mengistae, T. (2006). Investmentclimate and international integration. World Development, 34,14981516.

    Donner, J. (2004). Microentrepreneurs and mobiles: An exploration of theuses of mobile phones by small business owners in Rwanda. Informa-tion Technologies and International Development, 2, 121.

    Donner, J. (2006). The use of mobile phones by microentrepreneurs inKigali, Rwanda: Changes to social and business networks. InformationTechnologies and International Development, 3, 319.

    Donner, J., & Escobari, M. (2010). A review of evidence on mobile use bymicro and small enterprises in developing countries. Journal ofInternational Development, 22, 641658.

    Duncombe, R., & Heeks, R. (2002). Enterprise across the digital divide:Information systems and rural microenterprise in Botswana. Journal ofInternational Development, 14, 6174.

    ECLAC. (2011), ICT in Latin America, United Nations, Santiago, Chile.Esselaar, S., Stork, C., Ndiwalana, A., & Deen-Swarra, M. (2007). ICT

    usage and its impact on protability of SMEs in 13 African countries.Information Technologies and International Development, 4, 87100.

    Fagerberg, J. (1994). Technology and international dierences in growthrates. Journal of Economic Literature, 11471175.

    Fattouh, B., Scaramozzino, P., & Harris, L. (2005). Capital structure inSouth Korea: A quantile regression approach. Journal of DevelopmentEconomics, 76(1), 231250.

    Fernandes, A., & Paunov, C. (2012). Foreign direct investment in servicesand manufacturing productivity: Evidence for Chile. Journal ofDevelopment Economics, 97(2), 305321.

    Fisman, R., & Svensson, J. (2007). Are corruption and taxation reallyharmful to growth? Firm level evidence. Journal of DevelopmentEconomics, 83, 6375.

    Forman, C., Goldfarb, A., & Greenstein, S. (2014). Information technologyand the distribution of inventive activity. NBER working paper 20036.

    Forman, C., & Van Zeebroeck, N. (2012). From wires to partners: Howthe Internet has fostered R&D collaborations within rms. Manage-ment Science, 58(8), 15491568.

    Freeman, C., & Soete, L. (Eds.) (1997). The economics of industrialinnovation. Psychology Press.

    Friedman, T. L. (2005). The world is at: A brief history of the twenty-rstcentury. Farrar, Straus and Giroux.

    Girma, S. (2005). Absorptive capacity and productivity spillovers from

    LOPMENTGorg, H., & Greenaway, D. (2004). Much ado about nothing? Dodomestic rms really benet from foreign direct investment?. TheWorld Bank Research Observer, 19(2), 171197.

    Grossman, G. M., & Helpman, E. (1991). Innovation and growth in theworld economy. Cambridge: MIT Press.

  • Haskel, J. E., Pereira, S. C., & Slaughter, M. J. (2007). Does inwardforeign direct investment boost the productivity of domestic rms?.The Review of Economics and Statistics, 89(3), 482496.

    Hilbert, M. (2010). When is cheap, cheap enough to bridge the digitaldivide? Modeling income related structural challenges of technologydiusion in Latin America. World Development, 38(5), 756770.

    Howard, P., & Mazaheri, N. (2009). Telecommunications reform, Internetuse and mobile phone adoption in the developing world. WorldDevelopment, 37(7), 11591169.

    Hsieh, C.-T., & Klenow, P. J. (2009). Misallocation and manufacturingTFP in China and India. The Quarterly Journal of Economics, 124(4),14031448.

    Hu, A. G., Jeerson, G. H., & Jinchang, Q. (2005). R&D and technologytransfer: Firm-level evidence from Chinese industry. The Review ofEconomics and Statistics, 87(4), 780786.

    Indjikian, R., & Siegel, D. (2005). The impact of investment in IT oneconomic performance: Implications for developing countries. WorldDevelopment, 33(5), 681700.

    ITU. (2014). 2014 facts and gures. Accessed at: .Javorcik, B. S. (2004). Does foreign direct investment increase the

    productivity of domestic rms? In search of spillovers throughbackward linkages. The American Economic Review, 94(3), 605627.

    Jensen, R. (2007). The digital provide: Information (technology), marketperformance, and welfare in the south Indian sheries sector. TheQuarterly Journal of Economics, 122, 879924.

    Jorgenson, D. W. (2001). Information technology and the U.S. economy.The American Economic Review, 91(1), 132.

    Jorgenson, D. W., & Vu, K. (2005). Information technology and the worldeconomy. The Scandinavian Journal of Economics, 1074, 631650.

    Kaushik, P. D., & Singh, N. (2004). Information technology and broad-based development: Preliminary lessons from North India. WorldDevelopment, 32(4), 591607.

    Keller, W. (2004). International technology diusion. Journal of EconomicLiterature, 42, 752782.

    Klepper, S., & Simons, K. L. (2005). Industry shakeouts and technologicalchange. International Journal of Industrial Organization, 23(1), 2343.

    Koenker, R., & Bassett, G. (1978). Regression quantiles. Econometrica:journal of the Econometric Society, 3350.

    Kokko, A. (1994). Technology, market characteristics, and spillovers.Journal of Development Economics, 43(2), 279293.

    Kokko, A., Tansini, R., & Zejan, M. (1996). Local technologicalcapability and productivity spillovers from FDI in the Uruguayanmanufacturing sector. The Journal of Development Studies, 32(4),602611.

    Krugman, P. (1991). Increasing returns and economic geography. TheJournal of Political Economy, 99(3), 483499.

    Leamer, E. E., & Storper, M. (2001). The economic geography of theInternet age. NBER Working Paper No. 8450.

    Motohashi, K. (2008). IT, enterprise reform, and productivity in Chinesemanufacturing rms. Journal of Asian Economics, 19, 325333.

    Moulton, B. R. (1990). An illustration of a pitfall in estimating the eectsof aggregate variables on micro units. The Review of Economics andStatistics, 334338.Romer, P. (1986). Increasing returns and long-run growth. Journal ofPolitical Economy, 94(5), 10021037.

    Seker, M. (2012). Importing, exporting, and innovation in developingcountries. Review of International Economics, 20, 299314.

    Solow, R. (1987). Wed better watch out. New York Times Book Review,July 12.

    Spezia, V. (2011). Are ICT users more innovative? An analysis of ICT-enabled innovation in OECD Firms. OECD Journal: EconomicStudies, 99119.

    Stiroh, K. J. (2002). Information technology and the U.S. productivityrevival: What do the industry data say?. American Economic Review,92, 15591576.

    Tadesse, G., & Bahiigwa, G. (2015). Mobile phones and farmersmarketing decisions in Ethiopia. World Development, 68, 296307.

    UNCTAD. (2008). Measuring the impact of ICT use in business: The caseof manufacturing in Thailand, New York and Geneva.

    Von Hippel, E. (2005). Democratizing innovation. Boston: MIT Press.World Bank. (2006). Information and communication for development:

    Global trends and policies. Washington, DC.Yasar, M., & Morrison Paul, C. J. (2007). International linkages and

    productivity at the plant level: Foreign direct investment, exports,imports and licensing. Journal of International Economics, 71(2),373388.

    APPENDIXment. Oxford University Press.Paunov, C., & Rollo, V. (2015). Overcoming obstacles: The internetscontribution to rm development. World Bank Economic Review,Papers & Proceedings, 29(Suppl. 1), S192S204.

  • Table 10. Observations by country

    Country Observations Percentage sharein total

    Country Observations Percentage sharein total

    Country Observations Percentage sharein total

    Albania 199 0.40 The Gambia 153 0.31 Pakistan 843 1.69Angola 659 1.32 Georgia 243 0.49 Panama 587 1.17Antigua and Barbuda 116 0.23 Ghana 475 0.95 Paraguay 719 1.44Argentina 1,790 3.58 Grenada 129 0.26 Peru 1,464 2.93Armenia 262 0.52 Guatemala 858 1.72 Philippines 944 1.89Azerbaijan 291 0.58 Guinea 192 0.38 Poland 260 0.52The Bahamas 114 0.23 Guinea-Bissau 133 0.27 Romania 304 0.61Barbados 120 0.24 Guyana 136 0.27 Russian Federation 717 1.43Belarus 193 0.39 Honduras 595 1.19 Rwanda 183 0.37Belize 146 0.29 Hungary 248 0.50 Samoa 35 0.07Benin 90 0.18 Indonesia 1,122 2.24 Senegal 479 0.96Bhutan 215 0.43 Iraq 707 1.41 Serbia 327 0.65Bolivia 681 1.36 Jamaica 225 0.45 Sierra Leone 126 0.25Bosnia and Herzegovina 252 0.50 Kazakhstan 400 0.80 Slovak Republic 165 0.33Botswana 502 1.00 Kenya 636 1.27 Slovenia 243 0.49Brazil 1,077 2.15 Kosovo 200 0.40 South Africa 895 1.79Bulgaria 1,171 2.34 Kyrgyz Republic 154 0.31 Sri Lanka 462 0.92Burkina Faso 310 0.62 Laos 271 0.54 St. Kitts and Nevis 117 0.23Burundi 265 0.53 Latvia 211 0.42 St. Lucia 130 0.26Cameroon 320 0.64 Lesotho 88 0.18 St. Vincent and the Grenadines 129 0.26Cape Verde 96 0.19 Liberia 111 0.22 Suriname 152 0.30Central African Republic 135 0.27 Lithuania 209 0.42 Swaziland 259 0.52Chad 120 0.24 Macedonia 292 0.58 Tajikistan 247 0.49Chile 1,702 3.40 Madagascar 336 0.67 Tanzania 388 0.78Colombia 1,774 3.55 Malawi 83 0.17 Timor-Leste 82 0.16Democratic Republic of the Congo 517 1.03 Mali 654 1.31 Togo 102 0.20Republic of the Congo 91 0.18 Mauritania 214 0.43 Tonga 107 0.21Costa Rica 408 0.82 Mauritius 275 0.55 Trinidad and Tobago 308 0.62Ivory Coast 462 0.92 Mexico 2,454 4.91 Turkey 835 1.67Croatia 561 1.12 Micronesia 35 0.07 Uganda 515 1.03Czech Republic 165 0.33 Moldova 327 0.65 Ukraine 544 1.09Dominica 134 0.27 Mongolia 336 0.67 Uruguay 907 1.81Dominican Republic 289 0.58 Montenegro 60 0.12 Uzbekistan 320 0.64Ecuador 836 1.67 Mozambique 440 0.88 Vanuatu 81 0.16El Salvador 884 1.77 Namibia 307 0.61 Venezuela 158 0.32Eritrea 91 0.18 Nepal 328 0.66 Vietnam 953 1.91Estonia 232 0.46 Nicaragua 633 1.27 Yemen 300 0.60Fiji 47 0.09 Niger 85 0.17 Zambia 434 0.87Gabon 108 0.22 Nigeria 1,865 3.73 Zimbabwe 547 1.09

    604WORLD

    DEVELOPMENT

  • Table 11A. Description of variables used

    Name Description Mean Std. dev.

    Dependent variables

    Labor productivity Logarithm of the ratio of total annual sales over full time employment windsorized at the top and bottom 1% for anycountry-year, reported in thousand USD

    18 2.17

    Equipment investment Logarithm of the sum of 1 and the ratio of total annual expenditure for purchases of equipment over full timeemployment, reported in thousand USD

    2.1 1.08

    Certicates A dummy equal to one if the establishment has an internationally-recognized quality certication, such as ISO 9000or 14000 certications

    0.21

    Patents A dummy equal to one if the establishment has a registered patent and zero otherwise 0.39

    Industry Internet use

    Industry Internet use Percentage share of plants using email to communicate with clients and suppliers in industry j of country c in year t.Robustness tests include alternative measures for Internet (i) by industry, country-year and rm size, (ii) by industry,country-year and location type, (iii) by industry, country-year and geographic location and (iv) by country-year

    68.7 27.1

    Firm-level controls

    Employment Logarithm of the plants full-time employment 3.2 [25] 1.4Age Logarithm of the dierence between the year the survey was conducted and the year the plant was created 2.7 [15] 0.7Public ownership A dummy equal to one if the government or state own a share of 10% or more of the plant and zero otherwise 0.01Multi-plant rm A dummy equal to one if the plant belonged to a rm that had at least one other plant and zero otherwise 0.15Foreign ownership A dummy equal to one if the share of foreign ownership is bigger or equal to 10% and zero otherwise 0.12Exporter status An indicator that is equal to one if the plant exports (direct or indirect) 0.23Credit access Dummy variable is equal to one if the plant has a line of credit or loan from a nancial institution and zero otherwise 0.42Managerial expertise Logarithm of years of the managers experience 2.70 [15] 0.68Internet use Dummy variable where the plant has its own website and uses email and zero otherwise 0.41Industry A variable indicating in which sector the plant is operating: (i) food, (ii) wood and furniture, (iii) textiles, (iv)

    garments, (v) leather, (vi) non-metallic and plastic materials, (vii) chemicals and pharmaceuticals, (viii) electronics,(ix) metals and machinery, x) auto